# Agent handoff

A bounded transfer of task responsibility, state, evidence, and permitted actions from one agent or human role to another.

technical-term · Multi-agent & protocols · emerging · Reviewed 2026-08-09

## Definition

An agent handoff moves work between actors while preserving the goal, current state, source evidence, unresolved risks, authorization limits, expected output, and return or escalation path. A handoff transfers work—not accountability or standing beyond what governance has separately conferred.

Handoffs are supported by many frameworks; interoperable semantics and enterprise controls are still developing.

## Why it matters

Durable handoffs prevent the next actor from reconstructing critical state from prose alone.

The receiving agent must be independently authorized for the requested tools, data, and effects.

## System anatomy

- **Task contract:** Goal, scope, acceptance criteria, and deadline.
- **State packet:** Completed work, open issues, decisions, and evidence.
- **Authority check:** Validation that the recipient may accept and perform the work.
- **Return path:** Completion, rejection, correction, or escalation behavior.



## Important distinctions

- **Prompt forwarding:** A handoff carries structured state, evidence, and authority boundaries—not just another message.
- **Ownership transfer:** Business or lifecycle ownership requires explicit human authorization and recertification.



## Implementation signals

- Use typed handoff contracts
- Re-evaluate access for the receiving identity
- Record who transferred what and why

## Failure modes

- Context loss changes the task meaning
- Delegation silently expands tool or data access



## Related knowledge

- [Multi-agent system](https://aisdlc.ai/agentic-engineering/multi-agent-system) — A system in which multiple agents exchange work, state, or decisions under an explicit coordination and authority model.
- [Orchestrator or supervisor agent](https://aisdlc.ai/agentic-engineering/orchestrator-supervisor) — A coordinating component that decomposes work, routes tasks, monitors progress, and resolves or escalates outcomes across agents or tools.
- [Durable execution](https://aisdlc.ai/agentic-engineering/durable-execution) — Execution that persists state and can resume safely across process, model, context, worker, or environment interruptions.
- [Agent identity & delegated authority](https://aisdlc.ai/agentic-engineering/agent-identity) — The identity and authorization discipline that treats an enterprise agent as a non-human principal with attributable, purpose-bound, time-bound permissions.

## Sources and further study

- [A2A Project — Agent2Agent Protocol v1.0](https://a2a-protocol.org/latest/announcing-1.0/) — A protocol milestone for interoperable communication and collaboration between agents. **Use:** Standard or protocol. **Limitation:** A specification defines an interface or control pattern; conformance alone does not establish authorization, security, or fitness for a use case.
- [OpenAI — An open-source spec for Codex orchestration: Symphony](https://openai.com/index/open-source-codex-orchestration-symphony/) — An orchestration specification that connects project work, isolated agent execution, and reviewable delivery state. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.
- [Anthropic — Effective harnesses for long-running agents](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents) — Patterns for incremental progress and durable handoffs across context windows and extended execution. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.

---

This library synthesizes cited research, standards, official documentation, and clearly attributed practitioner perspectives. Maturity describes the state of a concept—not vendor endorsement, production readiness, or permission to deploy. Benchmarks and demonstrations do not replace use-case evaluation, governed controls, independent verification, or named human release authority.